Like many scientifically-inclined young men of his generation, Sokoloff chose to pursue a medical degree, although he wanted to do graduate work in physiology. In 1943, it was a practical decision: his undergraduate mentor, Lewis V. Heilbrunn, advised him that PhD students were likely to be drafted and forced to interrupt their work, while medical students could postpone their active military service and finish their degrees. Even before the war, job prospects were better for physicians than for scientific researchers; academic posts were limited, and the federal government had not yet established permanent programs to fund biomedical research. Thus, Sokoloff entered an accelerated three-year wartime medical program at the University of Pennsylvania. Accelerated programs were adopted by most American medical, dental, and nursing schools during the war, to provide health care personnel more rapidly to the armed forces. Most students were inducted into the Army Specialized Training Program, given the rank of Private First Class, and given basic military training on campus. The army covered all their tuition and lodging expenses, and paid them a small monthly stipend. In return, medical students were obligated to serve in the army Medical Corps for two years after they completed their internships and passed their licensing exams, even if the war concluded before that time. Apart from these wartime arrangements, their medical education was still administered by the universities.

Sokoloff found the early medical coursework dull, but persevered until he could begin the courses in biochemistry, physiology, and pharmacology, which he found more exciting. He received his MD in March 1946, and began his internship at Philadelphia General Hospital. His first rotation, in the psychiatric wards, was, he later recalled, "quite an eye-opener." The facility handled mainly acute psychiatric illness: schizophrenia, manic-depressive psychosis, depression, drug intoxication, and the many behavioral and neurological effects of chronic alcoholism and syphilis. The physical origins of many of these were obscure, and treatments were limited to electroconvulsive or insulin shock therapy, fever therapy, and psychotherapy. (The first anti-psychotic tranquilizers would not be available until the early 1950s.) Sokoloff was intrigued by his patients' strange behaviors, irrational thoughts, delusions, and hallucinations; they seemed real enough phenomena, though they defied attempts to explain them with physical science. (Indeed, mainstream American psychiatry was increasingly turning away from explanations based on possible physical or chemical mechanisms, and embracing psychoanalysis as a model.) When the war ended in 1945, the shift back to regular internship schedules meant that the wartime cohort's internship was extended by six months, and served in a single department. Sokoloff was assigned to psychiatry once again, and was able to develop further knowledge and competence in the field.

After his internship concluded in June 1947, Sokoloff passed his medical board exams and began his active army service. Because of his extended training in neuropsychiatry, the Surgeon General's office assigned him to direct the neuropsychiatry department at the Camp Lee, Virginia medical installation. For the next two years, he tended to a wide range of psychiatric and neurological problems presented by army personnel and their dependents, as well as practicing some internal medicine. He gave many of his patients the standard treatment of psychotherapy; it seemed to help some of them, but he still questioned the value and validity of the treatment. Neither his internship nor his army experiences with psychiatric patients had convinced him that psychiatry's descriptive diagnostic criteria and various modes of therapy had much rigorous scientific grounding. Mind and brain, he thought, had to be inextricably linked. He became increasingly interested in researching the physiologic and biochemical disturbances that might account for the signs and symptoms of mental illness.

In the late 1940s, biomedical scientists still knew relatively little about how the brain works. The brain, as Sokoloff often noted, performs an astonishing number of complicated tasks, more than any other organ in the body. Its diverse assembly of intricate parts mediate, regulate, or control many bodily processes. Its complexity alone makes it a challenge to study. But the brain also presents greater challenges to research than other organs: first, it is encased in the skull, so it is difficult to observe directly. Second, even if one could just look inside, the brain doesn't do its many tasks by moving things around, as other organs do, so observing it directly might not reveal much. How, then, could researchers begin to understand what goes on in the brain, both in normal conditions and when things go wrong? Nineteenth-century anatomists looked for evidence in postmortem dissections, correlating structural abnormalities (e.g. tumors) with the patients' clinical history of impaired neurological or behavioral function. By the early twentieth century, aided by x-ray imaging, neurologists were able to study large numbers of patients, many of them World War I veterans, with injuries or diseases in specific regions of the brain. By associating the specific sites of brain damage with a patient's particular impairment, they could begin to relate function and location within the brain. They also applied this approach to animal experiments--specific regions of the brain were either removed or destroyed, and the resulting impairments linked to the damaged sites. Alternatively, researchers could stimulate specific brain areas electrically or chemically, and observe the specific functional responses. Techniques for recording electrical activity in the brain, using probes, or near its surface using electroencephalography (EEG) sensors made it possible to make fairly precise maps of the topographic representations in regions of the brain for the various body parts. For example, the parts of the sensory and motor cerebral cortex representing the head, mouth, fingers, forelimbs, trunk, and lower limbs were mapped this way. But correlating brain region with sensory or motor function this way is slow work, and provides only a partial picture of what is happening. Likewise, physiologists and biochemists were only beginning to discover some of the many neurotransmitter substances and to understand how they function in the nervous system.

Ideally, researchers needed a way to determine, in an image of the whole brain, which specific areas were functioning (or not) during normal activities, and in abnormal conditions. In this way they might be able to "see," for example, what areas of the brain were affected by epilepsy or schizophrenia, where various drugs had their effects, what brain regions were active during sleep, and how the brain's activity changed from childhood to maturity to old age. A possible approach to this goal, suggested by British neurophysiologists Charles Sherrington and Charles Roy in 1890, would be to measure blood flow and indicators of energy metabolism (such as oxygen and carbon dioxide) going to and from the brain. Scientists knew that tissues that do physical work (muscle, kidney, or liver, for example) use energy in proportion to the amount of work done. When such tissues work harder, their energy metabolism (the biochemical reaction cycles that break down nutrients into energy and waste products) increases; they consume more oxygen and produce more carbon dioxide (CO2). The increased metabolism stimulates the circulatory system to deliver greater blood flow to supply the needed oxygen and carry off the CO2. Conversely, metabolic rate and blood flow decrease when the tissue is less active. It wasn't known for certain whether the functional activity of brain tissue was metabolically like the physical work done by muscles or kidneys, but if it were, it was likely that when part of the brain was more active, its metabolism and blood flow should increase, and vice versa.

The first step was to develop methods to measure blood flow in the whole brain. To measure oxygen and other indicators of metabolism, it is necessary to know how much blood is going through the tissue during the time in question. By the early 1940s, investigators had done quantitative studies of blood flow through the heart, lungs, liver, and kidneys. Carl F. Schmidt and Paul Dumke at the University of Pennsylvania had done the first quantitative cerebral blood flow measurement in rhesus monkeys. Using a bubble flowmeter surgically connected to the carotid arteries, their study measured normal cerebral blood flow, as well as the effects of various drugs, and the effects of different levels of oxygen deprivation on blood flow. But the flowmeter procedure was not suitable for use in humans--the surgery was invasive and risky, and there was no way to measure blood flow changes in conscious states, because the subjects had to be anesthetized (the animals used were sacrificed after the procedure.) By 1948, Schmidt and his colleague Seymour Kety had developed a better method, which could be used in conscious human subjects. They had their subjects inhale an inert gas, nitrous oxide (N2O), and monitored its concentration over time in the arterial blood and in one of the internal jugular veins. By applying a mathematical formula that related the arterial and venous N2O concentrations, and the known diffusion rate of N2O in brain tissue over time, Kety and Schmidt were able to calculate rates of cerebral blood flow (CBF), and measure metabolic indicators such as oxygen consumption and carbon dioxide production, in the brains of conscious, behaving humans and animals.

Sokoloff had studied pharmacology with Kety and Schmidt as a medical student, and was attracted to their rigorous scientific approach to diagnosis and therapy. Their nitrous oxide method impressed him as a powerful tool for studying psychiatric disorders, so he applied for a fellowship at Kety's lab when he finished his army service in 1949. He had, he admitted, "no skills, no methods, no brilliant research ideas," only a desire to work with the research group and learn as much as he could. Kety took him on, and during the next few years, Sokoloff learned how to work with radioisotope tracers, how to design and mathematically analyze kinetic physiological models, and the technical aspects of the procedure for the nitrous oxide method. Kety's team used the N2O method to learn about CBF in children, and about the effects of sleep, anesthesia, and drugs on CBF and oxygen consumption. They found that when consciousness was depressed (e.g. in coma or under anesthesia) oxygen consumption in the brain decreased. Curiously, they couldn't find a condition that showed increased oxygen uptake using the N2O method, and hoping to discover one, Sokoloff investigated the effects of hyperthyroidism (a disorder in which the thyroid gland produces too much of the hormone thyroxine) on CBF and energy metabolism. Hyperthyroidism was known to increase metabolism in other body tissues, and it also seemed to affect the brain, causing marked anxiety, so he and his colleagues expected that brain metabolism would be elevated as well. But it was not; cerebral energy metabolism remained normal in these patients. Why should this be? Reviewing the literature on thyroid hormones, Sokoloff could see that the effects of thyroxine on metabolism must be via stimulation of protein metabolism, i.e., the synthesis of protein molecules from amino acids, not the breakdown of glucose, as in energy metabolism. While thyroxine does stimulate protein metabolism and is required for normal growth and maturation of the developing brain, the adult brain has a fairly low rate of protein turnover, deriving much of its energy from glucose. There was no practical in vivo method for testing the effects of thyroxine on protein synthesis at the time, and Sokoloff lacked the biochemistry skills to do the work, but he would take up the problem several years later with great success.